Adaptive context-aware sensor array system for scene analysis in multiple human monitoring

Abstract

This paper presents a novel context-aware wireless sensor array system for scene analysis in multiple human monitoring. The goal of the research is to achieve real-time scene recognition and human identification through low-cost, low-data-throughput wireless pyroelectric sensor arrays. In the sensing information space, the scenes can be regarded as the combination of a bunch of basis in subspace. Such bases are extracted by using matrix factorization techniques. Geometry-regularized contextual pattern extraction and region of interest (RoI) identification schemes are developed and utilized. Experiment results demonstrate the capabilities of our context-aware sensing system on scene analysis and system adaption. The proposed context-aware wireless sensor system is not limited to pyroelectric sensors, it can be extended to various sensing modalities.